Tim, > > > ------------- Quote ------ > > > Date: Mon, 7 May 2007 10:29:50 -0700 > > > From: "John Mendenhall" <[EMAIL PROTECTED]> > > > To: "Artur Grabowski" <[EMAIL PROTECTED]> > > > CC: misc@openbsd.org > > > Subject: Re: new openbsd 4.0 server, panic on ufsdirhash > > > > > > Artur, > > > > > > We have done a forced fsck on the partition with the > > > error. The problem is, there is no data other than > > > the openbsd install. All I was trying to do was load > > > the source from the openbsd cd into /usr/src. > > > > > > I don't need to restore since this is a new machine. > > > I have not done anything to it. > > > > > > I'll just reinstall the entire thing. Unless someone > > > wants me to try something else. > > > > > > Thanks! > > > > > > JohnM > > > ------------------- /QUOTE > > > > > > John, > > > I've heard, and seen, a lot of odd problems that can't be > > duplicated > > > with the same error when there's either of the following true. > > > > > > 1) overclocked hardware > > > 2) bad system memory > > > > > > I'm doubting your system memory, but I'm curious about your > > > overclocking. > > > > > > I don't think I've followed very carefully what you've already > > tried, > > > and wonder if the mindset has ever drifted away from Hard Drives > > and > > > ATA controllers. > > > > > > Another thread suggested catting /dev/ad0s1 >/dev/null and seeing > > how > > > many errors you get. If you get errors, it might point to what > > can't > > > be read (and maybe can't be written then). You might have to use > > > another tool, but you should get the jist of what I'm trying to > > > suggest. > > > > All hardware is as received, no overclocking is being done. > > > > The system memory was the first issue we had. I have set > > the bios such that the system memory gives no errors on very > > long memtest runs. > > > > Currently, we are running a low level format of the two disks. > > No errors yet, but will run another day or so. > > > > Then, we'll reinstall the os and see how it goes. > > 'cat'ting the drive is simply reading data from the surface and sending > it to the bitbucket, so we can see if we can read the surface of the > drive without errors. > > A low-level format is an interesting twist, and I would like to see if > that helps. I've witnessed myself a drive "with bad blocks" dissapear > after a high-level format. It was the oddest of things, the FS itself > was corrupted and a disk check didn't help the situation. Maybe it was > a glitch, I don't know. I put that drive back into rotation.
We'll see how it goes. If I still get errors, I'll try to cat the drive to devnull and see what happens. It would be nice to get disk errors instead of a panic, though. Perhaps anything in a log file, or a console message. But, panic just stops everything and it's difficult to tell what actually happened. Or, perhaps, the drive is just going bad. I would have expected errors on installing the os if that were the case. Thanks! JohnM -- john mendenhall [EMAIL PROTECTED] surf utopia internet services
